Every LNA datasheet leads with a noise figure number, and every buyer's instinct is the same: lower is better, get the smallest number you can afford. That instinct isn't wrong, exactly. But it skips the question that actually determines whether a given NF matters at all: lower than what?
Noise figure only means something relative to the noise your system is already competing against. Below a certain frequency, your own environment (mostly man-made electrical noise, not the sky) is noisier than any amplifier you could buy, and chasing a 0.3 dB LNA to fight that is money spent on a problem you don't have.
Above that frequency — GPS, GNSS, most L-band work — there's no such floor to hide behind, and every tenth of a dB you leave on the table is signal you never recover. Same spec, opposite stakes, depending entirely on where you're operating.
This post is about setting the target before you go shopping, not after.
The number you're actually chasing isn't NF — it's sensitivity
Noise figure is an input to a bigger question: what's the weakest signal my system needs to reliably pull out of the noise? That's your sensitivity requirement, and it's set by your application, not your amplifier. A few examples of how differently that plays out:
- NOAA APT reception at 137 MHz — a satellite pass gives you a strong, predictable signal and a healthy link margin. Sensitivity isn't the bottleneck.
- GPS L1 at 1575.42 MHz — the signal arrives around -130 dBm, buried well below the noise floor before despreading even helps you. Every dB of system NF is a direct, linear hit to your acquisition and tracking margin.
- ADS-B at 1090 MHz — you're chasing weak, distant aircraft transponders and sitting next to strong nearby transmitters at the same time, so NF and dynamic range both matter, not NF alone.
Three receivers, three completely different relationships between "NF" and "does this actually work."
The part beginners skip: what's your noise floor without the LNA?
This is the single biggest lever in the whole decision, and it's the one most buying guides never mention. Every environment already has a noise floor before your receiver contributes anything — galactic background noise, atmospheric absorption, and man-made RF pollution. That floor isn't fixed; it changes dramatically with frequency, and at VHF it's usually the man-made component that dominates, not the sky.
Below roughly 200–300 MHz, external noise is often already higher than what a good LNA would add on its own — and the biggest contributor to that, outside of a genuinely remote location, is man-made noise: power lines, switching supplies, motors, and general electrical clutter. ITU-R Recommendation P.372 — the standard reference for background radio noise — gives an equivalent external noise figure for this at 137 MHz of roughly 8 dB in a rural environment, and 13–18 dB in residential or business/urban areas.
Compare that to a typical 1.0–1.5 dB LNA: the environment is contributing ten times or more the noise your amplifier is. In that world, dropping from a 1.5 dB LNA to a 0.5 dB LNA buys you almost nothing in real-world performance, because the system's noise floor was never set by your amplifier to begin with. It was set by the electrical noise around you, which no amount of LNA spec-chasing can filter out.
Above roughly 1 GHz, that external floor drops away, and the amplifier becomes the dominant noise source in the system. This is exactly why GPS and GNSS LNAs are built to extremely low NF (often 0.6 dB or better). At L-band, there's no atmospheric noise to hide behind, so the receiver's own contribution is the whole game.
The practical takeaway: check what band you're in before you shop by spec sheet. A low-VHF application and an L-band application should be sending you toward completely different price-to-performance tradeoffs, even if the marketing copy on both LNAs reads identically (e.g. "ultra-low noise").
What actually eats your noise figure before the LNA even gets a vote
One more lever worth flagging before you buy: the LNA's own NF is often not what determines your system's noise figure — whatever comes before it in the chain is. Cable runs and connectors ahead of the amplifier can quietly erase most of what a low-NF LNA was bought to deliver, which is the whole reason placement matters as much as the spec sheet.
We cover the mechanics in full in Model Your Full RF Receive Chain and System Noise Figure at the Antenna vs. Receiver — worth a read once you've settled on a target NF here.
Rule-of-thumb targets by application
Treat these as starting points for narrowing your search, not hard specs — your actual number should come from a real link budget for your system.
| Application | Typical band | External noise floor | Reasonable NF target |
|---|---|---|---|
| NOAA APT / weather satellite | 137 MHz | High (man-made-noise dominated) | 1.0–1.5 dB is usually plenty |
| 2m amateur band | 144–148 MHz | Moderate–high | 1.0–1.5 dB |
| ADS-B | 1090 MHz | Low | 0.8–1.2 dB, with dynamic range as a co-equal spec |
| GPS L1 | 1575.42 MHz | Low | 0.6 dB or better |
| GNSS multi-constellation | 1100–1700 MHz | Very low | 0.6–1.0 dB |
| Hydrogen line | 1420.4 MHz | Low | 0.6–1.0 dB, gain matters as much as NF here |
Notice the pattern: as center frequency climbs and the external noise floor drops away, the acceptable NF tightens.
A worked example: NOAA vs. GPS, same buyer instinct, opposite answer
Say you're comparing two 0.6 dB LNAs against two 1.5 dB LNAs, one pair for each application.
For NOAA APT at 137 MHz: unless you're in a genuinely remote, electrically quiet location, man-made noise alone puts the environment's effective noise figure somewhere around 8 dB (rural) to 13–18 dB (residential or urban) — well above any LNA on the market. The difference between a 0.6 dB and a 1.5 dB LNA is close to imperceptible in your decoded imagery, because the environment is contributing an order of magnitude more noise than either amplifier does. You'd be paying a premium for a spec your environment doesn't let you cash in. A mid-spec, lower-cost LNA is very often the correct engineering choice here, not a compromise.
For GPS L1: that same 0.9 dB gap is the difference between comfortable acquisition margin in a clear-sky rooftop install and marginal, unreliable lock under a tree canopy or near a window. Here, the 0.6 dB part isn't overkill — it's the whole reason the system works at all.
Same numbers, same "lower is better" instinct, opposite conclusion. The environment decided which one was right, not the spec sheet.
How to actually set your target
- Identify your weakest expected signal level for the application (from a link budget, a datasheet, or a known standard like GPS's -130 dBm nominal).
- Check your band's external noise floor. Below ~300 MHz, assume it's doing real work for you already. Above ~1 GHz, assume it isn't, and your LNA carries the full burden.
- Model your actual chain, cable loss and all, before you pick a part — a great LNA behind a long, lossy run can still leave you with a mediocre system NF. Our cascaded noise figure calculator does this math for you; plug in your real cable type and length rather than assuming it away.
- Leave margin. Real-world NF varies with temperature and unit-to-unit tolerance — don't design to a datasheet's best-case number and call it done.
- Buy to the target, not to the lowest number on the page. If your band already has a forgiving noise floor, that budget is better spent on gain, dynamic range, or filtering than on chasing another tenth of a dB you won't notice.
Where to go from here
Once you know your target NF, the next question is usually gain and placement — how much amplification you need, and where in the chain it needs to sit to actually deliver the number you calculated.
If you haven't already, run your specific setup through the cascaded noise figure calculator to see where your system currently stands, then browse GPIO Labs low-noise amplifiers filtered to the NF range your application actually needs — not the lowest number on the shelf.